Sulphur-2-LTX-2.3

🧪 Sulphur-2 Base (FP8, Distilled)

Model: SulphurAI/Sulphur-2-base

Format: FP8 (Quantized) Type: Distilled Base Model

📌 Overview

This release is an FP8-quantized version of the distilled Sulphur 2-base model, optimized for reduced VRAM usage and faster inference while maintaining strong output quality.

The model is intended for workflows that benefit from lightweight deployment, including ComfyUI pipelines, local inference setups, and experimentation with large-scale generative systems.

⚙️ Quantization

FP8 quantization was performed using:

👉 https://github.com/silveroxides/convert_to_quant

Special thanks to Silveroxides for providing the tooling that made this conversion possible.

🚀 Key Features ⚡ Reduced memory footprint (FP8) 🧠 Based on distilled Sulphur 2-base 🔄 Faster inference vs higher precision formats 🧩 Compatible with modern inference pipelines 📦 Use Cases Local image/video generation workflows Resource-constrained environments Rapid prototyping and testing ComfyUI / node-based pipelines ⚠️ Notes FP8 may introduce minor quality differences compared to FP16/FP32 Best results depend on sampler, scheduler, and pipeline tuning Ensure your environment supports FP8 inference 📜 License

Please refer to the original model page for licensing details: https://huggingface.co/SulphurAI/Sulphur-2-base